buf VS gRPC

The Buf CLI is a versatile tool designed for handling Protocol Buffers (Protobuf), a method of serializing structured data. It offers several key features, including managing Protobuf assets through the Buf Schema Registry (BSR), providing a linter to enforce optimal API design and structure, and a breaking change detector to maintain compatibility either in source code or at the wire level. Additionally, the Buf CLI includes a generator that activates plugins based on user-defined templates and a formatter to standardize the formatting of Protobuf files according to industry norms. It also integrates seamlessly with the Buf Schema Registry, supporting comprehensive dependency management.

We'll use the Buf CLI as an alternative to protoc so that we can save our generation configuration as YAML. Buf is compatible with protoc plugins.

Not so much anymore! Take a look at buf.build, it makes the whole thing notoriously easy :)

In my opinion it's a good idea, it's the architecture we use at work, and it works well for us. The main limitation to be aware of is that many PaaS don't support gRPC traffic (because of the proxies used). For example, DigitalOcean App Platform or Heroku if I remember correctly. If the way you want to host your backend is OK with HTTP/2 and gRPC traffic, then it's not a limitation. One way around this limitation is to use the gRPC-Web protocol, or the Connect protocol (https://connect.build/). Unfortunately, Dart's gRPC client does not support the gRPC-Web protocol outside the web platform. So for a mobile application, it's not usable at the moment. (If this PR were accepted, it would solve the issue: https://github.com/grpc/grpc-dart/pull/557.) As for Connect, no client is currently offered by Buf for Dart. Don't hesitate if you want to know more. That said, I'd advise you to use the Connect implementation for Go to implement your backend. Connect will enable your server to speak all three protocols (gRPC, gRPC-Web and Connect), which is very useful in the long term. What's more, the code is cleaner, and you benefit from official support for observability with OpenTelemetry. If you don't know Buf (the creators of Connect),I suggest you visit their website: https://buf.build/. :-) Good luck!

As mentioned in the intro, we are going to use Buf and Connect as our tools. We’ll start by installing the dependencies.

gRPC itself is quite nice, especially with buf which makes generating Go code much easier. The rest of the code was in a bad state. Unmaintained router packages, repository pattern without any actual benefit or a repository pattern.

The second big difference is that we now have auto-generated client and server stubs. For this task, I chose to use buf and the protobuf-ts plugin in order to generate idiomatic Typescript classes and objects. Not only do these classes describe the types we'll use in the server and client, but also includes the actual gRPC implementations used to serialize and send messages back and forth across the wire.

Our team has been using Buf (https://buf.build) recently, and they have a nice solution for schema dependency management.

I've found that https://buf.build/ is easier to use than protoc directly.

Consider checking out https://connect.build from https://buf.build. Supports a simpler protocol than grpc-web. Includes a js/ts client for frontend. Then you don’t necessarily need a rest layer, but could leverage the proxy your building.

gRPC, built on HTTP/2, inherently supports flow control. The server can push updates, but it must also respect flow control signals from the client, ensuring that it doesn't send data faster than what the client can handle.

Yes, grpc_cli tool uses essentially the same mechanism except implemented as a grpc service rather than as a stubby service. The basic principle of both is implementing the C++ proto library's DescriptorDatabase interface with cached recursive queries of (usually) the server's compiled in FileDescriptorProtos.

See also https://github.com/grpc/grpc/blob/master/doc/server-reflecti...

The primary difference between what grpc does and what stubby does is that grpc uses a stream to ensure that the reflection requests all go to the same server to avoid incompatible version skew and duplicate proto transmissions. With that said, in practice version skew is rarely a problem for grpc_cli style "issue a single RPC" usecases: even if requests do go to two or more different versions of a binary that might have incompatible proto graphs, it is very common for the request and response and RPC to all be in the same proto file so you only need to make one RPC in the first place unless you're using an extension mechanism like proto2 extensions or google.protobuf.Any.

While gRPC and Apache Thrift have served the microservice architecture well, CloudWeGo's advanced features and performance metrics set it apart as a promising open source solution for the future.

The loadBalancingConfig is what we use in order to decide which policy to go for (round_robin in this case). This JSON representation is based on a protobuf message, then why does the name resolver returns it in the JSON format? The main reason is that loadBalancingConfig is a oneof field inside the proto message and so it can not contain values unknown to the gRPC if used in the proto format. The JSON representation does not have this requirement so we can use a custom loadBalancingConfig .

The Dart implementation of gRPC which puts mobile and HTTP/2 first. It's built and maintained by the Dart team. gRPC is a high-performance RPC (remote procedure call) framework that is optimized for efficient data transfer.

gRPC is a high-performance, open-source RPC (Remote Procedure Call) framework initially developed by Google. It uses Protocol Buffers for serialization and supports bidirectional streaming.

In general, tunneling through HTTP2 turns out to be a great choice. There is a RPC protocol built on top of HTTP2: gRPC[1].

This is because HTTP2 is great at exploiting a TCP connection to transmit and receive multiple data structures concurrently - multiplexing.

There may not be a reason to use HTTP3 however, as QUIC already provides multiplexing.

I expect that in the future most communications will be over encrypted HTTP2 and QUIC simply because middleware creators can not resist to discriminate.

[1] <https://grpc.io>

Even in the https://grpc.io blog says this

gRPC